The Research On Personalized Recommendation Of The Mobile Life Services Based On LBS

The rapid development of mobile Internet provides a lot of Location BasedServices to users, which brings much convenience in the daily life. However, with theaddition of information, the overload problem makes the management and display ofthe importation difficult. Thus, it is an important and efficient way to deal with thisproblem by personal recommendation. It is not feasible to use traditionalrecommendation algorithm in the mobile Internet environment, because the mobileInternet has its unique characteristics: such as mobility, fast response and availability,location, correlation recognition. Thus, current research of personal recommendationin the mobile Internet environment of personalized recommendation is not enough. Inorder to improve the quality of services, some important questions, such as positionsensitive, sudden demand of users, and cold start, must be considered.According to the relevant requirements of the mobile Internet, this paper builds auser interest model based on LBS. This model can find the interests of users in thecurrent and next situation on the basis of foreseeing the next position of the users.Moreover, this model can give a complete service of mobile life recommended whichrecommends mobile life resource instances by building service resource directory.The main work includes:(1) building an ontology model of mobile life service. This work organizes thebasic concepts of domain knowledge, the concepts’ attributes and the hierarchy ofthem, etc.. Moreover, it combines life service concepts with interest concepts;(2) building a user interest model, namely the U-C-I model which combininguser’s position situation. It is a three-dimensional matrix, which recording the interestdegree of the interest category when a user is set in a specific location situation;(3)predicting the user’s next probably location which making therecommendation system no longer limited to the user the current position situation, soas to provide users more choices to meet their potential needs;(4) designing a four-steps recommendation mechanism, that is “the next position-interest-service resource directory-service resource instance”, meeting the demandof users to keep the diversity of recommended items. Among it, the interest-resourcedirectory service recommendation is based on the mobile life service ontology model, and the service directory-service resources recommendation is based on thesimilarity calculation between the users’ preference and the service instants.This paper which combining the location situation and the users’ interests cansatisfy the demand of location sensitive and users’ sudden needs. In addition therecommendation of the service resource instants doesn’t rely on the record of score ortransaction, and new users may obtain recommendations based on other users’characteristic under the same circumstance, so that the algorithm can alleviate theproblem of cold start efficiently. Based on the research of the recommendationprocess, the related algorithm, and the analysis of the test results, this work confirmsthe feasibility and effectiveness of the proposed algorithm, which can meet the users’demand of personalized recommendation under the mobile Internet.